The data on how much each congressman has voted with Trump comes from the website FiveThirtyEight. What follows isn't a criticism of the site's employees for collecting and disseminating this data. Instead I offer four related reasons that voters should not read too much importance into the percentages.
First: Republicans don't get high scores because they are slavish followers of Trump; he's the one following them. They're mostly voting according to the party's longstanding preferences, and he, partly under their influence, has gone along. That's why most Republicans have percentages in the 90s.
Second: Votes on fairly uncontroversial legislation boost the percentages for nearly everyone. The House vote on the Stop School Violence Act, a priority for the group Sandy Hook Promise, was 407-10. The scores went up for everyone in that majority, since Trump backed the bill.
Third: The scorecard has the potential to make bipartisanship on the part of Democrats look like weakness toward Trump. Forty-one House Democrats crossed the aisle to vote to let Obamacare's tax credits be used to defray premium costs for Cobra coverage, and 50 voted to expand Medicaid funding for the treatment of cocaine- and opioid-use disorders. Probably none of them (or the Republicans) voted that way because of pressure from Trump, or to signal their friendliness toward him.
The policies at issue in those votes have nothing to do with the reasons Trump inspires strong emotions, pro and con. If either vote had gone the other way, it would have inflicted very little political damage on the administration. Yet the scorecard could leave the impression that the Democrats who voted no are better members of the "resistance" than the ones who voted yes.
Fourth: The vote percentages are an extremely imperfect proxy for which congressmen are most and least helpful to Trump. Kyrsten Sinema is a House Democrat who is running for Senate from Arizona. Walter Jones is a conservative Republican from North Carolina. According to the scorecard, she is significantly Trumpier than he is. I'd say their respective scores tell you something about those two people, but more about the limits of this measure.
Among Republicans, the relative moderates are more likely to have high Trump scores than harder-right conservatives. (Which helps explain why so many of the moderates have been hit by the scorecard ads: They have high scores, and they're in swing districts where being seen as too tight with Trump hurts.)
The harder-right congressmen were more likely to vote against budget compromises and bipartisan initiatives. The 69 Republicans who voted against disaster relief for Puerto Rico all lowered their scores by doing so. In no important sense did their vote undermine their alliance with the president.
Rep. Thomas Massie, a conservative Republican from northeastern Kentucky, has been much less critical of Trump than Barbara Comstock has. But his pro-Trump score is much lower than hers - in part because he voted against the Puerto Rico bill and budget agreements, opposed a bipartisan bill to combat opioid abuse, and was one of the 10 votes against the school-safety bill.
Trump himself does not seem to be too upset about many of the votes "against him." Former GOP Rep. Ron DeSantis has fulsomely praised Trump and, in return, won the president's endorsement in his race to be governor of Florida. But dozens of Republicans in the House had more "pro-Trump" records than he did. Likewise, Rep. Matt Gaetz, another Floridian, regularly carries Trump's water on television, but he has one of the lowest pro-Trump scores of Republicans (although it's still 81.5 percent).
If you oppose Trump, you may reasonably wish to punish congressmen for being close allies of his. Put too much stock in these percentages, though, and you may punish the wrong things.